Release notes¶
v0.7.0¶
v0.7.0 adds bounded stream views and refines the browser interface.
- added stream publishing, live JSONL observation, pause/resume controls and stored-run browsing
- added bounded stream summaries, noteworthy observations and previous-visit comparisons
- expanded browser-native plotting with compact controls, a sidebar layout and explicit limit-recovery actions
- refined themes, view selection, filters, JSON defaults and content-loading feedback
v0.6.0¶
v0.6.0 makes replaceable live publishing bounded and inspectable.
- added
async_=Truetopublish_view()and@view(...)for attached and remote live publishing - added a bounded latest-wins
PublishWorker, with count and estimated-byte budgets - added
flush_views()and short publish-worker flushing duringstop_server() - exposed publish and storage queue counters through
/status - guarded the global in-memory view store against concurrent request/worker mutation
- bounded queued storage work by task count and estimated retained bytes
v0.5.1¶
v0.5.1 refines resource-safe watched-file serving and its defaults.
- file-backed CSV previews parse bounded rows and columns incrementally, avoiding unnecessary full-file scans and duplicate in-memory representations
- generated configuration now exposes watched-file materialization, uses a 10 MB automatic file-backing threshold, and enables latest-state persistence when storage is enabled
- generated configuration documents bounded async live-publish settings; synchronous publishing remains the compatibility default
- added repeatable release-gate benchmark coverage for watched-file settling, overload recovery, and publish-queue behaviour
- browser auto-refresh now coordinates requests, pauses in background tabs, and only reloads view-selector metadata when it changes
- browser assets are fingerprinted, packaged locally, and no longer require a runtime CDN; stable CLI, HTTP, watch-admission, and object-coercion concerns now have explicit module boundaries
v0.5.0¶
v0.5.0 is a perf-based update introducing bounded async live publishing and file-backed watched-file serving.
- added
async_=Truetopublish_view()and@view(...)for attached and remote live publishing - added a bounded latest-wins
PublishWorker, with count and estimated-byte budgets - guarded the global in-memory view store against concurrent request/worker mutation
- added memory-backed and file-backed watched-file materialisation
- file-backed CSV previews are parsed incrementally from disk instead of holding a raw byte window and then scanning the full file for a row count
- large watched files can now be represented by metadata and previewed from disk on demand
v0.4.0¶
v0.4.0 configuration clarity and watched-file reliability.
- simplified and clarified configuration structure and generated starter config
- improved watched-file behaviour, limits, freshness, and storage handling
- improved error reporting and visibility for failed publishes and watched-file updates
v0.3.1¶
v0.3.1 first plotsrv documentation release.
- added getting-started guides and configuration documentation
- added documentation for watched files, storage, freshness, and Python APIs
- improved examples and guidance for real-world usage
v0.3.0¶
v0.3.0 ease of use in real scripts and longer-running workflows.
- improved
@ps.view(...)andps.publish_view(...)as the main public API - added
launch_server=Truefor quick interactive use - improved publishing to an existing plotsrv server with
hostandport - added file-backed storage for latest views and snapshot history
- improved view discovery from Python code
- added storage CLI commands for inspecting and clearing stored outputs
v0.2.0¶
v0.2.0 API usability and project tidying.
- moved toward
@ps.view(...)andps.publish_view(...) - improved CLI behaviour
- improved config creation and population
- simplified the public API surface
- tidied internal structure ahead of broader documentation work
v0.1.0¶
- significant UI updates/rich features
Pre v0.1.0¶
Please see GitHub